Software and Implementation EnvironmentAll signal processing, feature extraction, model development, and statistical analyses were implemented using Python 3.11 through the Anaconda distribution and MATLAB R2024b.
ECG pre-processing, data handling, feature extraction, and numerical computations were performed using established Python scientific libraries, including NumPy, SciPy, pandas, WFDB, and BioSigKit.
The MATLAB Engine API for Python was used to call MATLAB R2024b routines from the Python environment when MATLAB-based signal processing or analytical functions were required.
Cross-validation, data partitioning, and performance evaluation were supported by scikit-learn, while statistical analyses and additional signal processing operations were performed using SciPy and MATLAB R2024b.
In this work, the authors used ChatGPT solely as a language editing tool, similar to grammar-checking software, to improve clarity, grammar, and sentence structure of the manuscript.
Competing interests
The authors declare no competing interests.
Software and Implementation Environment
All signal processing, feature extraction, model development, and statistical analyses were implemented using Python 3.11 through the Anaconda distribution and MATLAB R2024b. Python codes were developed in PyCharm 2023.2.4 on a Windows 10 Pro system equipped with an Intel® Core™ i7-4710HQ processor and 8 GB RAM, while computationally intensive deep learning experiments were also executed in Google Colab using a GPU-accelerated environment.
ECG pre-processing, data handling, feature extraction, and numerical computations were performed using established Python scientific libraries, including NumPy, SciPy, pandas, WFDB, and BioSigKit. The MATLAB Engine API for Python was used to call MATLAB R2024b routines from the Python environment when MATLAB-based signal processing or analytical functions were required. Deep learning architectures, including CNN, LSTM, GRU, Transformer-based models, and their hybrid variants, were implemented using TensorFlow/Keras. Cross-validation, data partitioning, and performance evaluation were supported by scikit-learn, while statistical analyses and additional signal processing operations were performed using SciPy and MATLAB R2024b. Figures and performance visualizations were generated using matplotlib.
MATLAB R2024b was used for the A-Test validation procedures and for the final HMM-based post-processing analyses, including the estimation of transition and emission probabilities and Viterbi decoding. This unified software configuration was used to maintain consistency across the analytical pipeline and to support reproducibility of the reported results.
Declarations of Generative AI and AI-assisted technologies in the writing process.
In this work, the authors used ChatGPT solely as a language editing tool, similar to grammar-checking software, to improve clarity, grammar, and sentence structure of the manuscript. No AI tools were used for generating scientific content, data analysis, or interpretation of results. All scientific content was independently developed, critically reviewed, and fully validated by the authors.
After using the tool, the authors carefully reviewed and edited all sections to ensure accuracy, scientific integrity, and compliance with journal standards, and take full responsibility for the content of the publication.
Consent to publish
All authors approve the manuscript and give their consent for submission and publication as open access.
